In this mini-series of tutorial, we will utilze Anthropic’s Model Context Protocol (MCP) as the foundation to build an AI agent that can trade foreign currencies based on real-time market intelligence. It will be divided into 3 installments:
Part 1: MCP primer and how to build your first MCP server
Part 2: Connecting to MCP servers for trading actions
Part 3: Putting everything together for an autonomous trading agent
Today, Inception Labs released the first commercially available Diffusion Large Language Model (dLLM) - Mercury Coder, and caused a big stir both in the research community as well as in the AI industry. In contrast to auto-regression LLMs (all the LLMs you know today), diffusion LLM works like your favorite AI image generators such as Stable Diffusion, where the final results emerge from a cloud of gibberish text. See one example below for the visualization of asking Mercury Coder to write a Python program to split an image into halves:
It is highly rewarding to see how AI “thinks” aloud step by step, trying alternatives, and self-validate and iterate, by running DeepSeek R1 locally on your computer instead of using their web/phone app. Here are a few examples that I find intriguing, during my two days of testing. It somehow reminds me of the writing style of Stephen Wolfram in many of his physics education series.
Running open-source AI models locally on our own computers gives us privacy, endless possibilities of tinkering, and freedom from large corporations. It is almost a matter of free speech.
For us GPU-poor, however, having our own AI computer seems to be a pricey dream. Macbook M3 Max? $3200, ouch! Nvidia 4090? $1850; that hurts even if you can get one. Microsoft Surface Laptop 6? Starting at $1200, still too much.
Everyone is GPU-poor these days, and some of us are poorer than others. So my mission is to fine-tune a LLaMA-2 model with only one GPU on Google Colab and run the trained model on my laptop using llama.cpp.
If you have a lot of private enterprise data, how can you use a ChatGPT-like AI system to help you search for relevant information faster? Or if you have lots of personal information, how can you find old memories without the exact keywords?
In this practical guide, I will show you 5 simple steps to implement semantic search with the help of LangChain, vector databases, and large language models. Best of all, I will use all open-source components that can be run locally on your own machine.
On February 24, 2023, Meta Research released LLaMA: a foundational, 65-billion-parameter large language model. Actually, there were three models: LLaMA 65B and LLaMA 33B were trained on 1.4 trillion tokens, while the smallest LLaMA 7B model was trained on one trillion tokens.
Then on March 13, 2023, a group of Stanford researchers released Alpaca 7B, a model fine-tuned from the LLaMA 7B model.
